Possible old and fixed bug in Postgres?
I was looking at perl CPAN Module (DateTime::Format::Pg) and saw that it did
something that seemed odd to me with time zones, based on the comment:
# For very early and late dates, PostgreSQL always returns times in
# UTC and does not tell us that it did so.
Early is before 1901-12-14 and late after 2038-01-18
A quick test setting my time zone to be America/Chicago I got
select '1900-01-01 00:00:00'::timestamptz;
timestamptz
------------------------
1900-01-01 00:00:00-06
(1 row)
and
select '2040-01-01 00:00:00'::timestamptz;
timestamptz
------------------------
2040-01-01 00:00:00-06
These seemed correct to me. I'm guessing this might have been a bug/feature of
pg in the long ago.
Steve
On 05/04/2023 11:18 CEST Steve Rogerson <steve.git@woodsideendurance.co.uk> wrote:
I was looking at perl CPAN Module (DateTime::Format::Pg) and saw that it did
something that seemed odd to me with time zones, based on the comment:# For very early and late dates, PostgreSQL always returns times in
# UTC and does not tell us that it did so.Early is before 1901-12-14 and late after 2038-01-18
A quick test setting my time zone to be America/Chicago I got
select '1900-01-01 00:00:00'::timestamptz;
timestamptz
------------------------
1900-01-01 00:00:00-06
(1 row)and
select '2040-01-01 00:00:00'::timestamptz;
timestamptz
------------------------
2040-01-01 00:00:00-06These seemed correct to me. I'm guessing this might have been a bug/feature of
pg in the long ago.
Judging by the commit message and changed test cases, probably:
--
Erik
Erik Wienhold <ewie@ewie.name> writes:
On 05/04/2023 11:18 CEST Steve Rogerson <steve.git@woodsideendurance.co.uk> wrote:
# For very early and late dates, PostgreSQL always returns times in
# UTC and does not tell us that it did so.
Early is before 1901-12-14 and late after 2038-01-18
...
These seemed correct to me. I'm guessing this might have been a bug/feature of
pg in the long ago.
Judging by the commit message and changed test cases, probably:
I think this was not fixed in full until 2008:
Either way, though, whatever Steve is looking at is far past its
sell-by date.
regards, tom lane
On 05/04/2023 11:23, Erik Wienhold wrote:
Judging by the commit message and changed test cases, probably:
That's the one. I can't see which pg version(s) this turned up in.
2023年4月6日(木) 0:02 Steve Rogerson <steve.git@woodsideendurance.co.uk>:
On 05/04/2023 11:23, Erik Wienhold wrote:
Judging by the commit message and changed test cases, probably:
That's the one. I can't see which pg version(s) this turned up in.
$ git tag --contains 921d749bd4c34c3349f1c254d5faa2f1cec03911 | head -1
REL8_0_0
Regards
Ian Barwick